This repository contains the Python code for training and evaluating a two-layer additive risk model on the Home Equity Line of Credit (HELOC) dataset used in the 2018 Explainable Machine Learning Challenge (https://community.fico.com/s/explainable-machine-learning-challenge). Our team (Chaofan Chen, Kangcheng Lin, Cynthia Rudin, Yaron Shaposhnik, Sijia Wang, and Tong Wang) won the FICO Recognition Award (https://www.fico.com/en/newsroom/fico-announces-winners-inaugural-xml-challenge).
 
If you plan to use our code, please cite our paper:

@article{chen2022holistic,
  title={A holistic approach to interpretability in financial lending: Models, visualizations, and summary-explanations},
  author={Chen, Chaofan and Lin, Kangcheng and Rudin, Cynthia and Shaposhnik, Yaron and Wang, Sijia and Wang, Tong},
  journal={Decision Support Systems},
  volume={152},
  pages={113647},
  year={2022},
  publisher={Elsevier}
}

You may find our paper here:

https://www.sciencedirect.com/science/article/abs/pii/S0167923621001573

If you would like to try out a trained model on a browser, you may use the following website that provides an interactive visualization of our model:

https://dukedatasciencefico.cs.duke.edu/models/